[lkml]   [2014]   [Mar]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH] [media] ov2640: add support for async device registration
Hi, Sylwester

On 3/20/2014 10:44 PM, Sylwester Nawrocki wrote:
> Hi Josh,
> On 19/03/14 10:17, Josh Wu wrote:
>> On 3/15/2014 5:17 AM, Sylwester Nawrocki wrote:
>>>> On 03/14/2014 11:12 AM, Josh Wu wrote:
>>>>>> + clk = v4l2_clk_get(&client->dev, "mclk");
>>>>>> + if (IS_ERR(clk))
>>>>>> + return -EPROBE_DEFER;
>>>> You should instead make it:
>>>> return PTR_ERR(clk);
>>>> But you will need this patch for that to work:
>>>> With this patch there is no need to overwrite any returned error
>>>> value with EPROBE_DEFER.
>> Thanks for the information. I will use this in v2 version.
> Oops, I missed somehow that it's v4l2_clk_get(), rather than
> clk_get(). So it seems it will not work when you return PTR_ERR(clk),
> since v4l2_clk_get() returns -ENODEV when clock is not found.

right, I missed that. So this version is still valid one. The v2 that I
already sent should be dropped.

> I think we should modify v4l2_clk_get() so it returns EPROBE_DEFER
> rather than ENODEV on error. I anticipate v4l2_clk_get() might be
> using clk_get() internally in future, and the v4l2 clk look up
> will be used as a fallback only. So sensor drivers should just
> do something like:
> clk = v4l2_clk_get(...);
> if (IS_ERR(clk))
> return PTR_ERR(clk);

I noticed that there are some driver like ov772x, ov9640 and etc, are
not supported the async probe yet.
If we return the EPROBE_DEFER for all v4l2_clk_get(), that will cause
the no-async probe function work incorrectly.
So IMO we can do your above suggestion after all sensors support async

> --
> Regards,
> Sylwester

Best Regards,
Josh Wu

 \ /
  Last update: 2014-03-26 08:01    [W:0.052 / U:36.896 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site